Re: patch to make --enable-cpp the default

[Zack Weinberg <zack at rabi dot columbia dot edu>]

On Wed, 05 May 1999 10:57:49 -0400, Dave Brolley wrote:
>The patch itself looks ok to me.....what I'm not clear on is whether
>the consensus is that this is a good idea at this time....

I will wait for an opinion from Jeff.

However, I should point out that if this isn't put in, 1.2 will have
no user-usable preprocessor, because of the patch

1999-04-06  Zack Weinberg  <zack@rabi.columbia.edu>

        * cppinit.c (install_predefs): Delete function.
        (cpp_start_read): Don't call install_predefs.
        (cpp_handle_option): Remove case 'u' and all refs to
        opts->inhibit_predefs.
        (print_help): Don't mention -undef.
        (initialize_builtins): Define __HAVE_BUILTIN_SETJMP__, to
        match cccp.
        * cpplib.h (struct cpp_options): Remove inhibit_predefs
        member.

        * cccp.c (predefs): Delete variable.
        (main): Remove case 'u' in argument parse loop,
        'inhibit_predefs' variable, and the code block that would
        process CPP_PREDEFINES.
        (initialize_builtins): Don't define __OBJC__, the driver will
        do that.

        * gcc.c (default_compilers): Remove -undef from all specs that
        invoke a C preprocessor.
        * ch/lang-specs.h: Likewise.
        * cp/lang-specs.h: Likewise.
        * f/lang-specs.h: Likewise.
        * objc/lang-specs.h: Likewise.

which I submitted under the assumption that --enable-cpp was already
the default, and which should not have gone in if we weren't going to
make it the default.  Right now, `ln -s PREFIX/lib/gcc-lib/T/V/cpp
/bin/cpp' gets you a /bin/cpp with no predefined macros at all and
which does not recognize -undef.  That will break Imake and probably
other things as well.  This is true for both cccp and cpplib.

For related issues see the "command-line discrepancies in cpp driver"
thread.

zw